Open billdenney opened 1 year ago
Adding the packages, that have been evaluated and are not of interest, somewhere in the source .md file is a good idea and would work well. I read in the .md file as part of the model training, and therefore this should be straightforward to implement.
We would then just need to announce this to the rest of the maintainers.
Hi, in the end it was decided that we should not put ignored packages in the TV source file, but instead to have an additional ignore
argument for CTVsuggest()
.
With a new install, you can now write:
CTVsuggest::CTVsuggest("Econometrics", ignore = c(
"GVARX" # better suited for TimeSeries TV
))
You should get the following output:
Econometrics Packages
JFE 0.9982844 JFE
iClick 0.9945763 iClick
lboxcox 0.9939188 lboxcox
wildrwolf 0.9934578 wildrwolf
gbmt 0.9932194 gbmt
Then for the automatic issue generation, I think this something that will be looked at later when more maintainers start to use CTVsuggest
. But in the mean time I can start working on it, so that when a decision is made about where the threshold information is stored, it can be implemented.
That makes sense. In that case, I'd suggest looking into a way to allow the maintainer to run a GitHub action on their repository on a schedule that would raise an error if there were un-addressed, suggested packages.
Yes, that makes sense thanks, the threshold can then be set in the action .yaml file.
I just ran
CTVsuggest(taskview = "Pharmacokinetics", n = 20)
, and it came up with a list of several useful packages. It also is suggesting some packages that are unrelated (no surprise and no issue there).I wonder if there could be some method of removing packages that have been evaluated and are not of interest from the suggested list.
My initial thought of a way to do this would be to put the list in the yaml header of the CTV source .md file like
But, I'm sure that there are many other ways that could also work.